Simulating a continous distribution

  • Context: Graduate 
  • Thread starter Thread starter MechatronO
  • Start date Start date
  • Tags Tags
    Distribution
Click For Summary
SUMMARY

This discussion focuses on simulating a continuous distribution using the Weibull distribution. The cumulative distribution function (CDF) F(x) is derived as F(x) = e^(-a) - e^(-a*x^b), and the transformation to obtain random values x is given by x = (-ln(e^(-a) - Y)/a)^(1/b). However, the user encounters an issue where no real solutions exist for Y > e^(-a), indicating a potential error in the formulation of F(x). The correct CDF must ensure that F(∞) equals 1.

PREREQUISITES
  • Understanding of cumulative distribution functions (CDF) and probability distributions
  • Familiarity with the Weibull distribution and its parameters (a, b)
  • Knowledge of logarithmic functions and their properties
  • Basic skills in random number generation and transformation techniques
NEXT STEPS
  • Review the properties of the Weibull distribution, focusing on its CDF and PDF
  • Learn about inverse transform sampling for generating random variables
  • Investigate the conditions under which CDFs are valid and ensure they approach 1 as x approaches infinity
  • Explore numerical methods for solving equations involving logarithmic and exponential functions
USEFUL FOR

Statisticians, data scientists, and software developers involved in simulations or modeling random variables according to specific distributions.

MechatronO
Messages
30
Reaction score
1
Say we want a set of random values that are distributed according to some distribution function f(x).

A common way to accomplish that is to find the cumulative distribution function F(x) for the distribution and then solve for x according to

F(x) = Y

x = F'(Y)

Then x will be distributed with the original distribution function, if F'(Y) is fed with random values Y ranging from 0-1.

I'm currently trying to do that with a weibull distribution

f(x) = a*b*xb-1*e-a*b*x^b

where F(x) should be

F(x) = e-a - e-a*x^b

when solving for x in F(x) I however get

x = ( - ln(e-a - Y)/a)1/b

When Y> e-a there are no real solutions. Is there a way to get around this? Have I done something wrong?
 
Physics news on Phys.org
Your F(x) can't be right. F(∞) should be 1.
 

Similar threads

  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 1 ·
Replies
1
Views
441
  • · Replies 3 ·
Replies
3
Views
3K
  • · Replies 8 ·
Replies
8
Views
2K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 5 ·
Replies
5
Views
3K
  • · Replies 1 ·
Replies
1
Views
1K
Replies
4
Views
2K
  • · Replies 6 ·
Replies
6
Views
5K
  • · Replies 14 ·
Replies
14
Views
2K